{"about":{"non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","site":"https://codewithpapers.app","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page","syntology":{"site":"https://syntology.ai","developers":"https://syntology.ai/developers","mcp":{"server":"https://syntology.ai/mcp","transport":"streamable-http","server_card":"https://syntology.ai/.well-known/mcp/server-card.json","auth":{"type":"trial token, no account","trial_token":"https://syntology.ai/api/oauth/trial/token","method":"POST","docs":"https://syntology.ai/developers"}},"have":"https://syntology.ai/api/graph/have?x=<method, arXiv id or title> (free, answers coverage only)","paper_base":"https://syntology.ai/paper/","atlas_base":"https://app.syntology.ai/?focus="},"machine_readable":[{"url":"https://codewithpapers.app/llms.txt","what":"the machine catalog: every machine-readable file, counted"},{"url":"https://codewithpapers.app/index/manifest.json","what":"paper-to-code index by arXiv id, with Syntology's counts"},{"url":"https://codewithpapers.app/search/manifest.json","what":"site search index (titles, authors) and its files"},{"url":"https://codewithpapers.app/download","what":"bulk files: Syntology's layer, described there"},{"url":"https://codewithpapers.app/build_manifest.json","what":"the build record: inputs, counts, exclusions, probes"}]},"url":"/paper/the-information-pathways-hypothesis","title":"The Information Pathways Hypothesis: Transformers are Dynamic Self-Ensembles","arxiv_id":"2306.01705","date":"2023-06-02","proceeding":null,"authors":["Md Shamim Hussain","Mohammed J. Zaki","Dharmashankar Subramanian"],"abstract":"Transformers use the dense self-attention mechanism which gives a lot of flexibility for long-range connectivity. Over multiple layers of a deep transformer, the number of possible connectivity patterns increases exponentially. However, very few of these contribute to the performance of the network, and even fewer are essential. We hypothesize that there are sparsely connected sub-networks within a transformer, called information pathways which can be trained independently. However, the dynamic (i.e., input-dependent) nature of these pathways makes it difficult to prune dense self-attention during training. But the overall distribution of these pathways is often predictable. We take advantage of this fact to propose Stochastically Subsampled self-Attention (SSA) - a general-purpose training strategy for transformers that can reduce both the memory and computational cost of self-attention by 4 to 8 times during training while also serving as a regularization method - improving generalization over dense training. We show that an ensemble of sub-models can be formed from the subsampled pathways within a network, which can achieve better performance than its densely attended counterpart. We perform experiments on a variety of NLP, computer vision and graph learning tasks in both generative and discriminative settings to provide empirical evidence for our claims and show the effectiveness of the proposed method.","url_abs":"https://arxiv.org/abs/2306.01705v1","url_pdf":"https://arxiv.org/pdf/2306.01705v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"the-information-pathways-hypothesis","repo_url":"https://github.com/shamim-hussain/ssa","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"graph-learning","task_name":"Graph Learning"},{"task_slug":"graph-regression","task_name":"Graph Regression"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/graph-regression-on-pcqm4mv2-lsc","task":"Graph Regression","dataset":"PCQM4Mv2-LSC","model":"EGT+SSA+Self-ensemble","rank_in_archive_order":14,"of":20,"metrics":{"Validation MAE":"0.0865"},"uses_additional_data":false},{"leaderboard":"/sota/graph-regression-on-pcqm4mv2-lsc","task":"Graph Regression","dataset":"PCQM4Mv2-LSC","model":"EGT+SSA","rank_in_archive_order":16,"of":20,"metrics":{"Validation MAE":"0.0876"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-imagenet","task":"Image Classification","dataset":"ImageNet","model":"Swin-T+SSA","rank_in_archive_order":601,"of":1060,"metrics":{"Top 1 Accuracy":"81.89%"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-wikitext-103","task":"Language Modelling","dataset":"WikiText-103","model":"Transformer+SSA+Self-ensemble","rank_in_archive_order":23,"of":89,"metrics":{"Test perplexity":"17.18","Validation perplexity":"16.54"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-wikitext-103","task":"Language Modelling","dataset":"WikiText-103","model":"Transformer+SSA","rank_in_archive_order":27,"of":89,"metrics":{"Test perplexity":"17.60","Validation perplexity":"16.91"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-enwiki8","task":"Language Modelling","dataset":"enwik8","model":"Transformer+SSA","rank_in_archive_order":22,"of":42,"metrics":{"Bit per Character (BPC)":"1.024"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2306.01705","atlas_url":"https://app.syntology.ai/?focus=2306.01705","mcp":null,"developers":"https://syntology.ai/developers","held":true,"reason":"arxiv_id_in_atlas","checked_against":{"atlas_ids":549449,"atlas_ids_sha256":"afd72399dafd9906e9062927f70655952385a3d75488558e1f0484e950ab09b7","read_at":"2026-10-05T01:46:21Z"},"agent_calls":[{"tool":"get_citation_path","arguments":{"paper_1":"2306.01705"},"arguments_in":{"paper_2":"another paper's arXiv id or title"},"call":"get_citation_path(paper_1=\"2306.01705\", paper_2=\"…\")"},{"tool":"get_concepts_for_paper","arguments":{"arxiv_id":"2306.01705"},"call":"get_concepts_for_paper(arxiv_id=\"2306.01705\")"}],"by_repo":[{"repo":"shamim-hussain/ssa","kind":"official","samples":0,"ran":0,"instrument":0,"unverified":0}],"by_repo_is":"one row per repository in the archive's code links, Syntology's graph links or the samples; samples 0 means no sample is linked to this paper, a statement about Syntology's coverage, not about the repository: with harvested_for_other_papers present, Syntology harvested the repository for that many other papers, and without it, Syntology harvested nothing from it; placed_by_identical_code counts samples whose link records no repository, placed here because identical code was harvested from this repository for another paper; ran counts executions on a synthesized input and instrument counts failures of Syntology's instrument, not of the code"},"code_links_note":null,"arxiv_metadata":null,"syntology_extracted_results":null,"record_sha256":"43457d5aa8a7ce90e5b7d133753c494fefcd07cbc4fbdef0312ed4f9d4ca2719","record_changed_at":"2026-09-28","record_changed_at_basis":"first_hashed"}